GI Forum, Nov 10, 2020
Earth-related coordinates of spatial data can be either
Geometrical computations on projected or geocentric coordinates can be done using Euclidean geometry, where all lines are staight.
Spherical (or ellipsoidal) geometry computes distances, direction, areas etc. over the surface of the sphere (ellipsoid).
I believe that applying Eucledian geometry to geographical coordinates is the number one most common error made in spatial data science (closely followed by ignoring the support of data).
technological advances have made spherical geometry a good option, but inertia in legacy GIS and data science languages have slowed down its adoption / uptake
We have gotten used to Plate Carree, and are OK with it.
Even GeoJSON has written down that the world is 2D.
Compared to other projections:
Compared to an ellipsoid, or sphere:
Consider, for a moment, the difference between: